Hackers use AntiFox AI agents in AWS credential-theft attacks

Hackers are leveraging the AntiFox AI platform, hosted on Alibaba Cloud, to carry out attacks aimed at stealing AWS credentials, according to Arctic Wolf. This development is significant as it highlights a sophisticated method of exploiting cloud services to gain unauthorized access to sensitive information. The attackers reportedly use AntiFox, an open-source penetration testing tool, to perform malicious activities such as vulnerability exploitation and AWS secret harvesting.

AntiFox, initially derived from ARTEX 0.1.1, is an open-source tool with a Chinese-language interface that facilitates the deployment of AI agents to automate penetration-testing tasks. These tasks include mapping exposed services and probing target endpoints for vulnerabilities. Arctic Wolf observed that attackers hosted AntiFox on Alibaba Cloud to exploit this functionality. The malicious infrastructure allowed them to deploy webshells, validate credentials, and discover configurations in targeted environments. By using valid AWS identities, including IAM users and EC2 roles, the attackers could make authenticated requests to Amazon Web Services APIs.

One notable incident involved the deployment of an ASPX webshell on an internet-exposed IIS server. This allowed the attackers to execute operating system commands, access application settings, and retrieve sensitive data such as database connection strings. They also copied configuration and log files to a web-accessible directory. In another case, the attackers used valid IAM credentials and EC2 roles to probe AWS endpoints, cycling through multiple identities and making hundreds of API requests within a short time frame. Although many attempts were blocked, some succeeded, allowing the theft of production secrets.

The implications of these attacks are substantial for businesses relying on AWS and other cloud services. Unauthorized access to AWS environments can lead to data breaches, financial loss, and reputational damage. Arctic Wolf's report provides guidance for defenders, recommending the correlation of IIS webshell activity with unusual AWS access and the review of successful API calls within bursts of denied attempts. They also advise revoking compromised AWS sessions and rotating exposed credentials to mitigate these threats.

The AntiFox findings come in the wake of similar attacks involving ARTEX AI, another penetration-testing framework developed in China. ARTEX went closed-source after being used in attacks on South Korean banks. It was also linked to a cyberespionage campaign by the Chinese group Vapor Panda, which targeted over 200 organizations in 14 countries. Despite the risks, AntiFox remains available on GitHub, with its publishers prohibiting its use against live systems or for illegal activities, though no technical controls enforce these restrictions.